.png){kind=icon}
.png){kind=icon}
Return
An efficient and robust two-phase two-point material point method for fluid-solid interaction problems
.png){kind=icon}
PRE
AI
DOI：10.1016/j.compgeo.2026.108091.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Numerical simulation of large-scale landslide-induced surge disasters often faces significant challenges, including complex fluid–solid interactions, fluid volumetric locking, non-physical clustering of particles, and high computational costs. To address these issues, this study proposes an efficient and robust two-phase two-point Material Point Method (TT-MPM) based on mixture theory. Two key algorithms are implemented to enhance the robustness of the method: (1) A projection method, designed to effectively alleviate non-physical stress oscillations induced by volumetric locking; (2) A particle shifting technique based on a spatial hash grid (PST-SHG), designed to proactively optimize particle distribution and reduce macroscopic volume errors caused by non-physical clustering of particles. The proposed framework is implemented using the Taichi language for high-performance Graphics Processing Unit (GPU) parallel computing. Three numerical examples—three-dimensional (3D) saturated soil, seepage through porous dam, and onshore landslide and induced surge—demonstrate that the simulation results are consistent with both theoretical solutions and experimental data. Meanwhile, compared with the traditional Material Point Method, the proposed method has significantly suppressed non-physical stress oscillations, enhanced the uniformity of particle distribution, and effectively ensured the macroscopic volume conservation of fluids. By leveraging Compute Unified Device Architecture (CUDA) cores, the proposed GPU parallel program achieves a 14.6 speedup compared to traditional serial Central Processing Unit (CPU) computing. This study provides an accurate, stable, and efficient computational tool, offering robust support for the prediction and analysis of largescale geological hazards.
Keywords:
Two-phase two-point Material Point Method
fluid-solid interaction
volumetric locking
particle shifting technique
GPU parallel computing